The short message service is an important service among mobile communication services, the income thereof accounting for a growing percentage of total operator income. Hence, the operating security and stability of the short message center, the core of what provides the service, have also become a focus of increasing attention of operators and device manufacturers.
At present, there are mainly two technical schemes used by short message service, which are respectively: double-network double-plane and cluster.
Wherein, in the double-network double-plane scheme, two or more short message offices are connected to network by the way of pairwise interconnecting, and once one short message office is faulted, the other short message office may take over all the service; at the same time, the traffic is transferred from an over loaded short message center to a light loaded short message center to balance the traffic. This double-network double-plane technology improves the security and stability of short message center from disaster-tolerating and load-balancing point of view, but has the following disadvantages:
Firstly, when one short message office is faulted, all the services can only be handed over to the other office, and if the other office is also faulted or overloaded, the service will break down or be affected.
Secondly, the double-network double-plane technology requires short message office joining therein to be pairwise inter-networked. To enhance the security of the short message center, multiple short message offices are formed into a multi-network multi-plane structure, however, this networking way is very inconvenient. Because, at first, each office should record the information of all the other office, and when one office is added, deleted or modified, the data of all the other office needs to be reconfigured; next, given that there are N office in network, N*(N−1)/2 links need to be maintained, and the more the office are, the more complex the links are, and the more difficult the management for the links is, thus the higher the probability of faults occurring is.
Thirdly, there lacks the function of monitoring and management for the traffic load of whole system. Each node manages the traffic relevant to itself in a distributed way, without one node being able to monitor the traffic of whole network, incapable of configuring and performing a flexible load-balancing policy.
A cluster system equips each short message office with one cluster crunode server, and then manages the cluster crunode server by a cluster dispatch center, thereby implementing uniform management for two or more short message offices, also, functions as disaster-tolerating and shunting. However, the cluster system also has some disadvantages:
Firstly, its architecture is much complicated, one cluster crunode server being also needed to be built for every short message office aside from the cluster dispatch center being needed to be built.
Secondly, aside from supporting the star structure centering around the cluster dispatch center, the cluster system also supports the way of cluster crunode servers connecting to one another directly, and the disadvantages of this way are similar to the double-network double-plane structure.
Thirdly, the cluster dispatch center has only one single service processing module, without being divided into a signaling processing module, a service processing module and a short message peer to peer protocol agent (SMPP AGENT) module, only providing uniform message interface. Thus, the cluster crunode server needs to convert messages received from these three modules using different protocols into one uniform message format when exchanging messages with the cluster dispatch center, and also needs to convert protocols for messages when sending messages to these three modules. These conversions increase the burden of hardware and the time of processing messages.
Fourthly, the operation status and load condition of the short message center is reported regularly to the cluster dispatch center by cluster crunode, and the cluster dispatch center then invokes a load-balancing policy to shunt the load, and the real time property of load reporting is not strong; meanwhile, the accuracy of load reporting is also not good due to different operating systems and the processing mechanism of the short message center and the like factors.